<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Mirjana Barovic on Medium]]></title>
        <description><![CDATA[Stories by Mirjana Barovic on Medium]]></description>
        <link>https://medium.com/@m.barovic?source=rss-f1a4240c1c38------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*dmbNkD5D-u45r44go_cf0g.png</url>
            <title>Stories by Mirjana Barovic on Medium</title>
            <link>https://medium.com/@m.barovic?source=rss-f1a4240c1c38------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 17:21:25 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@m.barovic/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[A Back-and-Forth With AI to Shape Better Conversations]]></title>
            <link>https://medium.com/comsystoreply/a-back-and-forth-with-ai-to-shape-better-conversations-0ebd5bb7c8cc?source=rss-f1a4240c1c38------2</link>
            <guid isPermaLink="false">https://medium.com/p/0ebd5bb7c8cc</guid>
            <category><![CDATA[lego]]></category>
            <category><![CDATA[better-conversations]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Mirjana Barovic]]></dc:creator>
            <pubDate>Wed, 29 Apr 2026 12:12:01 GMT</pubDate>
            <atom:updated>2026-04-29T12:12:01.345Z</atom:updated>
            <content:encoded><![CDATA[<p><em>How I used AI to tighten the structure of a 60‑minute LEGO® Serious Play session (and what the final agenda looked like)</em></p><figure><img alt="AI generated clock of small bricks as a model on a base plate" src="https://cdn-images-1.medium.com/max/1024/1*t6k3lW8VKFKN2R2pWUgALg.png" /><figcaption>AI generated clock of small bricks as a model on a base plate</figcaption></figure><p>I held a 60 minute session <strong>on challenging situations in facilitation delivered via LEGO® Serious Play (LSP) </strong>for a group where I didn’t even know who had used LSP before. That’s a lot of ambition for a very small timebox.</p><p>This happened during <a href="https://medium.com/comsystoreply/getting-better-at-facilitating-conversations-f345fdc604ca">Advanced Facilitation Training with Mirna</a> (Trainer’s Toolbox) in London. Tiny group, lots of practice, lots of feedback. Which meant: no hiding behind theory. The session had to work in the room, with real people and real timing.</p><p>So I used AI as a sparring partner. Not to generate a “perfect agenda”, but to help me iterate faster, stress-test transitions, tighten questions, and make the painful cuts. All that without spamming colleagues with version 17 of my draft.</p><p><strong>What you’ll get in this post:</strong></p><ul><li>The final 60‑minute agenda (copy/paste friendly)</li><li>Five examples of where AI helped (and where it didn’t)</li><li>The one thing I’ll reuse everywhere: a simple two‑column flipchart table that made learning visible</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*k9sffOIfSlxmKGfg1xvJJA.png" /></figure><h3>If you take only one thing from this story…</h3><p>The <a href="#5655"><strong>two-column flipchart table</strong></a> should be it. It’s the bit that made the learning visible.</p><h3>Jump to</h3><ul><li><a href="#e55b"><strong>Where AI might support in design</strong></a></li><li><a href="#ea5b"><strong>The back‑and‑forth: 5 things I did with AI</strong></a></li><li><a href="#5655"><strong>If you only copy one thing</strong></a></li><li><a href="#6095"><strong>The final 60‑minute agenda</strong></a></li><li><a href="#7e4d"><strong>What I learned</strong></a></li><li><a href="#4742"><strong>Try it</strong></a></li></ul><h3>Where AI might support in design</h3><p>Through my years of working as Scrum Master, PO, PMO, and Team Coach, I’ve designed a lot of meetings and workshops. When I got stuck either during design or during facilitation, it was usually because of timeboxing, or because I’d prepared too many options “just in case”.</p><p>Most of the time I learned to adjust on the spot, roll with it, and hope nobody noticed the spike in my stress level.</p><p>Then I signed up for <a href="https://trainers-toolbox.com/advanced-facilitation-training/">Advanced Facilitation Training</a> with Mirna in London, and we had to design and run real segments in a tiny group. Suddenly, my usual “let’s roll with it” approach felt a bit too optimistic.</p><p>So I did what I’ve started doing more and more lately: I brought <strong>AI into the design phase</strong>. Not to generate a perfect agenda, but…</p><figure><img alt="Text in image: to give me something I could argue with." src="https://cdn-images-1.medium.com/max/1024/1*vZH3PXf0hyhvuZBTIe-csA.png" /><figcaption>Text: to give me something I could argue with</figcaption></figure><p>Something that would keep asking: “Okay, but what happens at minute 27 when people get stuck?” And, very practically, <strong>something that would let me iterate through multiple versions without spamming colleagues</strong> with draft after draft.</p><h3>My AI journey started with a simple problem: time</h3><p>I was the facilitator of a 60 minute session on challenging situations in facilitation, using LEGO® Serious Play (LSP). The topic was big, the method can be deep, and the time slot was… not.</p><p>On top of that, I didn’t know what was the groups experience with LSP. Perhaps some had worked with it before, perhaps some didn’t. And this made the design problem even more interesting.</p><p>My first AI prompt was basically:</p><blockquote><em>“Help me design a 50‑minute session. It should be hands‑on, include LSP, and land in concrete takeaways.”</em></blockquote><p>The first versions of the agenda I got back looked neat. They also looked like they had been <em>designed for a world where humans don’t need instructions, transitions, or time to think</em>.</p><p><strong>That’s when the real value kicked in: the back-and-forth.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m_024CWWB4RwqA3lRcW1jA.png" /></figure><h3>Here’s the honest version of my interaction with it: Five things I did with AI</h3><h3>1) I used it to generate variants, not answers</h3><p><strong>What I asked AI for:</strong> Three different session flows:</p><ul><li><strong>Go deep quickly</strong> (assumes LSP experience)</li><li><strong>Teach the method first</strong> (assumes beginners)</li><li><strong>Hybrid</strong> (tries to respect both)</li></ul><p><strong>What it gave me:</strong> Clean agendas that looked reasonable on paper and slightly delusional in a 60‑minute slot.</p><p><strong>What I kept/changed:</strong> I kept the hybrid direction, because the group was mixed.</p><p><strong>I changed my own intentions and expectations from “agenda perfection” to “what will actually survive in the room”</strong>.</p><p>Seeing the three variants side by side made the trade‑offs visible: more time on skill‑building meant less time for shared sense‑making, and vice versa.</p><h3>2) I used it to stress-test transitions (my usual weak spot)</h3><p>My biggest design misgiving wasn’t the exercise itself. It was the <strong>movement between parts</strong>:</p><ul><li>How do I get from warm‑up to “real scenario” without it feeling random?</li><li>How do I keep it practical, not just “nice builds”?</li><li>How do I converge without crushing the group’s ownership?</li></ul><p><strong>What I asked AI for:</strong> Transition scripts and signposts: how to move from warm‑up to scenario, and from building to debriefing.</p><p><strong>What it gave me:</strong> Lots of options. Many too wordy, but some genuinely helpful.</p><p><strong>What I kept/changed:</strong> I kept the structure of a few signposts and rewrote them into sentences I could actually say. How did I check if AI’s answers were what I was looking for? Simple, with one question:</p><figure><img alt="Text in image: Does this leave time for people to think and speak, or does it just give me more to explain?" src="https://cdn-images-1.medium.com/max/1024/1*mYmM_gVDFT0TiD9a0SHD2g.png" /><figcaption>Text: Does this leave time for people to think and speak, or does it just give me more to explain?</figcaption></figure><h3>3) I used it to refine questions</h3><p><strong>What I asked AI for: </strong>I iterated with AI on prompts like:</p><ul><li>“Any thoughts?”</li><li>“Does that make sense?”</li><li>“Any questions?”</li></ul><p>(My issue with these is that they are unspecific and yes/no oriented).</p><p><strong>What it gave me:</strong> A long list of good questions.</p><p><strong>What I kept/changed: </strong>I kept a tiny set I could remember during the session while listening to the group discussion:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pmf8_OTmCoHXTNXI1p-CgQ.png" /></figure><p>I ended up with a small set of sharper prompts I could actually remember and use while listening, instead of a long list that would keep me in my head.</p><h3>4) I used it to make a hard call: what to cut</h3><p>There was a point where the agenda still had too many moving parts. One of them was the <a href="https://www.liberatingstructures.com/13-wise-crowds/">Wise Crowds exercise from Liberating Structures</a> (an exercise to instantly engage a group in helping one another).</p><p><strong>What I asked AI for:</strong> Ways to include the exercise without adding time.</p><p><strong>What it gave me:</strong> Clever compressions that still relied on humans moving at 1.5x speed.</p><p><strong>What I kept/changed:</strong> <strong>I skipped the Wise Crowds exercise entirely.</strong> We didn’t even have time to properly introduce it, and I didn’t want a half‑baked version that confused people.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UXe8EmUrTo9f8dVQXkytyw.png" /></figure><p>Instead, I moved that time into what I actually wanted:</p><ul><li>more time on discussing the <strong>shared model</strong></li><li>more time on creating a clear <strong>converging tool.</strong></li></ul><p><strong>That decision felt like facilitation maturity</strong>. AI helped me see the trade‑offs faster, but <em>I was still the one making the final design decisions</em>.</p><h3>5) Rename sections so I couldn’t hide behind vagueness</h3><p><strong>What I asked AI for:</strong> Better section names that describe what’s supposed to happen.</p><p><strong>What it gave me:</strong> Names that were more explicit than my original “discussion” and “wrap-up”.</p><figure><img alt="Examples of changes I made with AI as helper tool. From Introduction to Opening and Frame; from Closing to Harvest and insights." src="https://cdn-images-1.medium.com/max/820/1*5HuBuXGe11dXDwk4qdP-Dg.png" /><figcaption>Examples of changes I made with AI as helper tool</figcaption></figure><p><strong>What I kept / changed:</strong> I kept the forcing function. Renaming “discussion” to <em>“</em>share + converge on a flipchart table<em>”</em> made it harder for me to drift. <strong>It also made it clearer what success should look like.</strong></p><p>These wording tweaks later made it easier to stay on track in the room.</p><h3>The final result: my 60-minute LEGO® Serious Play session design</h3><p>This is what the tightened structure landed on (in plain English).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Vp7yOkYJbkLQe7oEomR0XA.png" /></figure><p><strong>Materials:</strong> LEGO bricks, one base plate (or shared surface), flipchart + marker, timer.</p><h3>Starting the session: Quick welcome &amp; frame (2–3 min)</h3><p>Introduction into what we’re exploring: difficult facilitation situations, and practical moves to handle them.</p><p><em>Success looks like:</em> everyone knows the point of the session and the timebox.</p><h3>1) Short LSP warm-up (10–12 min)</h3><p>Enough building/sharing to get newcomers comfortable without losing experienced folks.</p><p><em>Success looks like:</em> everyone has built and spoken at least once.</p><h3>2) Shared scenario build (15–20 min)</h3><p>Build together around a challenging facilitation situation (the messy middle — also known as the Groan Zone — shows up fast here, which is kind of the point).</p><p><em>Success looks like:</em> a shared model that represents a real challenge, not a vague metaphor.</p><h3>3) Share + converge on a flipchart table (20–25 min)</h3><p>As people explained the shared model, I stayed in “curious facilitator” mode and filled a simple two‑column table:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z4dfrcrvSvWoYaP5HzErIA.png" /></figure><h3>If you want to copy one thing, this should be it.</h3><h4>This became the <strong>converging tool</strong>.</h4><p>This table mattered more than it seems. While I listened and wrote, I also asked follow‑up questions and repeatedly pointed the group back to the flipchart:<em> “Which of these items I wrote down would you phrase differently?”. </em>This way<em> </em>we could validate the wording together and gradually steer toward an ending that still respected the timebox.</p><h4>It was also deliberately <strong>simple</strong>.</h4><p>In a short slot, I needed generous insights and practical solutions, not a tool that demanded my attention. The table let me focus on what people were actually saying (and what they weren’t saying yet), instead of juggling something more complex just to make it look fancy.</p><p>In the debrief, some participants said they really valued seeing their ideas made visible and organised in real time.</p><p><em>Success looks like:</em> a table full of clear, usable moves.</p><h3>4) Close (2–3 min)</h3><p>Quick verbal round: one concrete action per person.</p><p><em>Success looks like:</em> each person leaves with one sentence they could actually try next week.</p><p>No Wise Crowds exercise. No extra conceptual detours with bricks. <strong>Just enough structure to make the learning visible.</strong></p><h3>What I learned beyond “AI can write agendas” (and what I’m taking into my job)</h3><p>AI helped me iterate faster, tighten transitions, and arrive with sharper prompts. It also helped me notice my favourite facilitation anti‑pattern: <em>adding “just one more” exercise.</em></p><p>It also made it easier to iterate without bothering a friend with 17 versions of the same agenda (which is… considerate of me).</p><p>What it didn’t do: read the room, sense energy, notice who went quiet, or decide when to slow down. In other words, it didn’t do the facilitation. It just improved my odds of showing up with a plan that left me enough attention to actually facilitate.</p><p>Would the session have worked without AI? Probably.</p><p>Would I have got to this version, this fast, with this much clarity on the trade‑offs? Probably not.</p><p>And to be clear: the best part of the session wasn’t the AI support. It was the moment the group looked at the flipchart table and went, “Yes, that’s exactly it.” That’s the point of facilitation: making thinking visible, together.</p><h3>Try it out</h3><p>Pick one meeting you run often (retro, planning, training segment).</p><ul><li>Ask AI for <strong>three alternative flows</strong></li><li>Ask for <strong>five sharper debrief questions</strong></li><li>Ask: “What if everyone were a beginner?” and “What if everyone were experienced?” Then design your own middle ground, if the group is mixed</li><li>Then do the human part: choose what fits your group, cut what doesn’t, and leave space for people to think</li></ul><p>And don’t stop at the first iteration. Give feedback to AI on <em>why</em> something works or doesn’t work in your setting. You’ll get better answers.</p><p>AI won’t facilitate the conversation for you.</p><p>But it can help you show up with a cleaner plan, and more attention left for the room.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0ebd5bb7c8cc" width="1" height="1" alt=""><hr><p><a href="https://medium.com/comsystoreply/a-back-and-forth-with-ai-to-shape-better-conversations-0ebd5bb7c8cc">A Back-and-Forth With AI to Shape Better Conversations</a> was originally published in <a href="https://medium.com/comsystoreply">comsystoreply</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Getting Better at Facilitating Conversations]]></title>
            <link>https://medium.com/comsystoreply/getting-better-at-facilitating-conversations-f345fdc604ca?source=rss-f1a4240c1c38------2</link>
            <guid isPermaLink="false">https://medium.com/p/f345fdc604ca</guid>
            <category><![CDATA[facilitation]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[lego-serious-play]]></category>
            <category><![CDATA[education]]></category>
            <dc:creator><![CDATA[Mirjana Barovic]]></dc:creator>
            <pubDate>Fri, 27 Mar 2026 12:28:02 GMT</pubDate>
            <atom:updated>2026-03-27T12:28:02.212Z</atom:updated>
            <content:encoded><![CDATA[<h4><em>Advanced facilitation, LEGO® Serious Play and a little help from AI</em></h4><figure><img alt="The image shows multiple people around a table building models with LEGO bricks and writing notes on stickies" src="https://cdn-images-1.medium.com/max/1024/1*gNjBBIYolYp6VehJG1sUYw.png" /><figcaption>A preview of what it might look like when LSP is involved in a meeting</figcaption></figure><p>I spend a lot of my working life in rooms (physical or virtual) where people are trying to do good work together, often under slightly messy conditions. And what I’m really doing in those rooms isn’t “running workshops” so much as <em>facilitating conversations</em>: the ones that happen in meetings, retrospectives, trainings, planning sessions, or whenever a group needs to think (and decide) together.</p><p>A team with no previous Scrum Master. <br>A legacy product that comes with ten years of history and three years of “we should really fix this.” <br>A small new-product setup where I’m wearing the Product Owner hat and thinking about UX.</p><p>In all of those contexts, better conversations mean better work.</p><p>That’s exactly why I didn’t want to coast on <em>“I’ve facilitated a lot, so I’m fine.”</em> I wanted sharper tools and better instincts, especially for the messy middle of a conversation, when a group gets stuck or too polite or too chaotic. So I joined the <a href="https://trainers-toolbox.com/advanced-facilitation-training/"><strong>Advanced Facilitation Training</strong></a> by <strong>Mirna</strong> (Trainer’s Toolbox) in London.</p><h3>The setup was almost suspiciously intense…</h3><figure><img alt="The Image shows multiple combined pictures from the onsite training. The people are discussing different topics in a group setting — sitting in a circle with a flipchart next to them." src="https://cdn-images-1.medium.com/max/1024/1*v-WmGQ7NxtVxXAig9hKX0g.png" /><figcaption>The group from our Advanced Facilitation Training in London</figcaption></figure><p>… in a good way: four participants plus Mirna, lots of practice, lots of feedback, and a tone that was kind, playful and high‑energy. We worked with ideas like the <strong>facilitation diamond</strong> (diverge → groan zone → converge), <strong>powerful questions</strong>, <strong>decent debriefs</strong> (the part I often under-design), and the less talked‑about things: voice, body, room setup.</p><p>Outside the training space, I had my own little rituals, which were part recovery, part sense-making. The venue coffee was… optimistic, so after lunch we went out for real coffee, because apparently I’m the kind of facilitator who can handle a groan zone but not bad espresso.</p><p>And because the days were intense in that <em>good, my-brain-is-full</em> way, I needed something that wasn’t more input. So I’d hop on the 109 double-decker bus and let London roll by. And one evening I went to see Wicked<strong>, </strong>where I fully switched off, which I can recommend as a professional development strategy with excellent singing.</p><p>All of that “off time” turned out to be more useful than it sounds…</p><h3>Because it’s when I started seeing the real challenges of my own training design.</h3><figure><img alt="A group discussion displayed as models on a base plate" src="https://cdn-images-1.medium.com/max/1024/1*DUTQjXpY4L6QneyxlFFjjA.png" /><figcaption>A group discussion displayed as models on a base plate</figcaption></figure><h3>Challenging Situations in Facilitation: 60 minutes of LEGO® Serious Play</h3><p>One of my biggest design challenges was a 60‑minute LEGO® Serious Play (LSP) block. I had to create a practical and reflective space <em>to explore the difficult situations facilitators face, and share strategies for handling them.</em> The design itself was up to me. Naturally, I picked the most difficult method I could come up with to facilitate this topic in such a short amount of time: LEGO® bricks.</p><p>The tricky part: I didn’t know the other participants yet, so I genuinely couldn’t tell whether I’d be facilitating for LSP veterans, complete newcomers, or that very realistic mix of <em>“I’ve heard of it and once touched a brick in 2019.”</em></p><p>That uncertainty created a classic dilemma:</p><ul><li>Do I optimise for speed and depth, and risk losing people who don’t have the basics?</li><li>Or do I teach the foundations at the risk of boring the experienced folks and spending half the slot on “how to LSP” instead of using it to get to the actual point?</li></ul><p>Either way, time was the real enemy. If I assumed the participants had enough experience and skip the precursory exercises, we might not use LSP properly and miss the gist of what the models can surface. On the other hand, if I focused too much on introducing LSP as a method to them, I’d burn precious minutes just trying to get everyone onto the same page.</p><h3>What I actually did</h3><p>I chose a middle path: a short skills warm‑up to get everyone building and sharing, and then a proper shared scenario build.</p><p>When people presented their shared model, I stayed in “curious facilitator” mode by asking questions, checking I’d understood, and capturing the insights on a flipchart in two columns:</p><figure><img alt="The image shows a simple table with the title “Lego Serious Play — Challenging Situations in Facilitation”. The table presents two columns. Right-side title “What the challenge was”; left-side title “How to resolve it”." src="https://cdn-images-1.medium.com/max/1024/1*J8Ch7g-az1GAgAaVTBQBuA.jpeg" /><figcaption>A preview of the table as a converging tool on which I captured what was said regarding challenges and resolutions</figcaption></figure><p>Simple structure, yet surprisingly effective.</p><p><em>The flipchart became our converging tool</em>: something the group could literally point at, refine, and build on — while I tried to write things down as clearly as possible.</p><p>The feedback reflected the trade‑off I was aiming for. The experienced LSP folks said we could probably have jumped straight into the scenario. The newcomers, on the other hand, appreciated the warm‑up and said the timing felt balanced. Which I’ll happily count as a small win.</p><p>And the notes on the flipchart were appreciated too, because they made the learning visible: not just “nice builds”, but <em>concrete challenges and practical ways to handle them</em>.</p><h3>Where AI helped (and where it didn’t)</h3><p>And yes, <strong>AI</strong> was part of this too.</p><figure><img alt="A table that presents examples of where AI support helped me tighten my structure and refine the questions I asked. For example, refining the questions “What is the groan zone like for you” into “What are the signs we’re in the groan zone?”" src="https://cdn-images-1.medium.com/max/1024/1*k5l3lwnDnETX_Srsg45lQg.jpeg" /><figcaption>A few examples where AI support helped me sharpen my questions and design a strong agenda</figcaption></figure><p>I used it as a beginner design partner: to test a few agenda variants, tighten the flow, and iterate on titles and questions (especially the ones aimed at experienced facilitators). It widened the option space.</p><p>But the final calls on what to keep, what to cut, what I felt comfortable doing live, <em>stayed very much my own decisions</em>.</p><h3>What I’m taking back to my day job</h3><p>My main takeaway wasn’t “new techniques,” but <em>a shift in attention</em>: designing better conversations in workshops and meetings means planning for the messy middle, matching the first exercise to the real purpose and tone of a meeting, and taking debriefs seriously.</p><h4>Try it!</h4><p>If you design workshops in agile/product/tech contexts: maybe pick one meeting you run a lot — and redesign just the opening and the debrief. Small changes do make a big difference.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f345fdc604ca" width="1" height="1" alt=""><hr><p><a href="https://medium.com/comsystoreply/getting-better-at-facilitating-conversations-f345fdc604ca">Getting Better at Facilitating Conversations</a> was originally published in <a href="https://medium.com/comsystoreply">comsystoreply</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>